Family-Based CBT Beats Active Control in Paediatric OCD — But the Margin Is Smaller Than We Like to Say
- Single-centre RCT, n=130 children and adolescents (8–17 years, 52.3% female, mean age 13.3, mean CY-BOCS 25.8) randomised 1:1 to 14 sessions of family-based CBT with ERP (FCBT, n=64) versus family-based psychoeducation and relaxation training (FPRT, n=66).
- End-of-treatment CY-BOCS was significantly lower for FCBT than FPRT: 15.9 (SD=8.7) vs 19.9 (SD=8.1); between-group estimate −3.89, 95%CI [−6.83, −0.96], p=0.01, Cohen's d=0.47, 95%CI [0.09, 0.85].
- The −3.89-point advantage fell **below** the predefined minimal clinically important difference of 4 points — the trial met statistical but not clinical significance on the primary endpoint.
- Adverse events on the Negative Effects Questionnaire (NEQ-20) showed no significant group differences; dropout was higher in FPRT (12 vs 4) — exposure was not less tolerable than psychoeducation.
For two decades the field has cited paediatric OCD CBT effect sizes that come almost entirely from waitlist-controlled trials. TECTO is one of the few large RCTs to compare exposure-based CBT against an active psychological control delivered with the same intensity, by the same kind of clinicians, with the same family involvement. The outcome is sobering: ERP wins, but by less than a Y-BOCS notch.
What the data actually shows
The Copenhagen team designed TECTO specifically to neutralise the inflation that comes from comparing a structured therapy against "treatment as usual" or a waiting list. Both arms got 14 weekly family-based sessions; both arms had homework; both arms had warm therapists. The only systematic difference was whether the child was guided into exposure with response prevention or guided into relaxation and education about anxiety.
Mean CY-BOCS dropped by ~9.9 points in the FCBT arm (25.8 → 15.9) and by ~5.9 points in FPRT (25.8 → 19.9). Both arms moved. ERP moved further. But the 4-point gap that would have crossed the prespecified MCID was not reached — the upper bound of the confidence interval (−0.96) just barely excluded zero.
There are two honest readings. The first: ERP's true edge over a credible non-exposure intervention is real but modest, around d=0.47, which is roughly half of what the older waitlist-controlled meta-analyses suggested. The second: families and clinicians who cannot access trained ERP providers — a chronic problem outside university clinics — can still get clinically meaningful improvement (mean −5.9 CY-BOCS points) from a structured family programme that any well-supervised therapist can deliver.
For your practice
Two implications, both immediate.
First, when you sit with a family weighing a 14-week ERP programme against the next-best available local option, the conversation changes. The honest framing is no longer "ERP works, the alternative does not" — it is "ERP gives a moderate additional benefit on top of a structured non-exposure programme, with comparable side-effect burden, and slightly better treatment completion." For families terrified of exposure, that framing is more accurate and more respectful.
Second, the modest effect size is a flag for our protocols, not for ERP itself. TECTO used 14 sessions across 16 weeks. Concentrated formats (Bergen B4DT, intensive 5-day delivery) consistently report larger effects in adult and adolescent OCD. If your standard outpatient model shows the TECTO-sized gap, the question is whether dose, density, or family integration is the limiting factor — not whether ERP "works".
ERP beats a credible psychological placebo in paediatric OCD — but by less than a Y-BOCS notch, and the families with the worst access to it should hear that honestly.
Single-centre Danish trial; participants and therapists could not be masked; primary outcome relied on assessor-rated CY-BOCS with some self-report missingness; FPRT was unusually well-designed and may have been more therapeutically active than the typical "active control" in older paediatric OCD trials.